Efficient Time-Stepping for Numerical Integration Using Reinforcement Learning

DOI10.1137/21M1412682zbMATH Open1520.65050arXiv2104.03562OpenAlexW4367185279MaRDI QIDQ6039265FDOQ6039265

Sebastian Peitz, Karlson Pfannschmidt, Michael Dellnitz, Marvin Lücke, Eyke Hüllermeier, Christian Offen, Sina Ober-Blöbaum

Publication date: 4 May 2023

Published in: SIAM Journal on Scientific Computing (Search for Journal in Brave)

Abstract: Many problems in science and engineering require an efficient numerical approximation of integrals or solutions to differential equations. For systems with rapidly changing dynamics, an equidistant discretization is often inadvisable as it either results in prohibitively large errors or computational effort. To this end, adaptive schemes, such as solvers based on Runge--Kutta pairs, have been developed which adapt the step size based on local error estimations at each step. While the classical schemes apply very generally and are highly efficient on regular systems, they can behave sub-optimal when an inefficient step rejection mechanism is triggered by structurally complex systems such as chaotic systems. To overcome these issues, we propose a method to tailor numerical schemes to the problem class at hand. This is achieved by combining simple, classical quadrature rules or ODE solvers with data-driven time-stepping controllers. Compared with learning solution operators to ODEs directly, it generalises better to unseen initial data as our approach employs classical numerical schemes as base methods. At the same time it can make use of identified structures of a problem class and, therefore, outperforms state-of-the-art adaptive schemes. Several examples demonstrate superior efficiency. Source code is available at https://github.com/lueckem/quadrature-ML.

Full work available at URL: https://arxiv.org/abs/2104.03562

zbMATH Keywords

machine learningquadratureinitial value problemsreinforcement learningtime-stepping

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Numerical quadrature and cubature formulas (65D32) Initial value problems, existence, uniqueness, continuous dependence and continuation of solutions to ordinary differential equations (34A12) Multistep, Runge-Kutta and extrapolation methods for ordinary differential equations (65L06) Numerical methods for initial value problems involving ordinary differential equations (65L05)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Vectorized adaptive quadrature in MATLAB
• The MATLAB ODE Suite
• DGM: a deep learning algorithm for solving partial differential equations
• \({\mathcal Q}\)-learning
• Solving Ordinary Differential Equations I
• Geometric Numerical Integration
• A family of embedded Runge-Kutta formulae
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Machine learning for fast and reliable solution of time-dependent differential equations
• Data-driven approximation of the Koopman generator: model reduction, system identification, and control
• A Fourier Integrator for the Cubic Nonlinear Schrödinger Equation with Rough Initial Data

Cited In (1)

• Learning Robust Marking Policies for Adaptive Mesh Refinement

This page was built for publication: Efficient Time-Stepping for Numerical Integration Using Reinforcement Learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6039265)